Analyzing Large Groups of Specific Users

[czmudzin]

I need to analyze the behavior of a specific set of unique visitors. To do this, I need to narrow my analysis to 2,500 unique visitor IDs. My only thought is to create 5 sets of segments, using the 'Equals any of' operator with the visitor ID values, and export the data from Data Warehouse. Is there a better way to approach this (outside of using clickstream data from a data feed)?

[Jennifer_Dungan]

With that many users, I would be tempted to use a classification...

Basically set up a "classification" that will hold one value, something like "true" or maybe "include". Then use the classification importer with those specific ids to set the value.... 

Basically, then you would only have to create a segment for "classification equals true" or  "classification equals include"

Its sort of a cheat, and using classifications in a non-standard way, but it's a lot easier than trying to create a segment with that many operators (particularly since the equals any of is a bit of a pain to work with when dealing with a lot of values. Also, I think there are limitations to have much you can add)

[czmudzin]

@Jennifer_Dungan This is a great idea, and I didn't consider using classifications. One concern that I have is that a lot of IDs fall into 'Unspecified' because there's so many. If they're in a classification, would they all show up properly in analysis workspace?

[Jennifer_Dungan]

They show up as "unspecified" or in the "(Low Traffic)" bucket? Unspecified means the value isn't set (this should never occur due to there being "so many"), low traffic on the other hand just means the reporting suite has hit the unique value threshold for the month (I think that classification would bypass that, since it's processing the data on the server, it's not trying to render it.... but I've never tried on a low threshold value before)

Anyway, it should be a fairly quick implementation, and if it doesn't work then we might be back to square one... but if we're talking about "low traffic" then the segment definitely won't work.... 

[czmudzin]

@Jennifer_Dungan Sorry, I did mean "Low Traffic," not "Unspecified." I don't know how Analysis Workspace would handle low traffic values if they're classified. Perhaps I will give it a shot and let you know what the outcome is!

[Jennifer_Dungan]

Hahah no worries, I figured, but I wanted to confirm there wasn't another layer of "issue" that we had to deal with.   Based on your description, I assumed "low traffic".

Let me know how the classification works for you, fingers crossed.

[czmudzin]

@Jennifer_Dungan I just chatted with our developer on this. He previously applied a classification to 'low traffic' values in Workspace and it did not render all of the data. Just thought I would share this piece of knowledge.

I think my approach will be creating 5-6 segments and extracting the data from data warehouse and working in XLS.

[Jennifer_Dungan]

Unfortunately, segments don't work with Low Traffic either... If something is in "Low Traffic" the segment won't return those matches...

I've tried that in Workspace.. known items that I had in Low Traffic that a segment should have picked up... no "low traffic" bucket showed when the segment was applied... I don't know if it will work differently on a Data Warehouse export.

May I ask approximately how many User Ids you have in your dimension? A few months ago Adobe upped the Low Traffic thresholds to 2 million unique values... are you sure that you will still be impacted by low traffic?

[czmudzin]

It is my understanding that DWH exports all data, regardless of traffic volume. That's why I was going create a few segments, apply them each a single DWH report and export 5-6 different files. 

We have about 1.5M IDs each month, but I am only interested in 2.5K of them.

[Jennifer_Dungan]

DWH will export all data, but if its constrained by a segment, it only exports data that falls into that segment... if the segment doesn't work with your low traffic, then your Data Warehouse exports will still miss that data.

If you only have about 1.5M ID, you should be able to pull this without the Warehouse, or if your analysis needs to be done in XLS, the classification based solution would still be easier to make the segment for pulling data...

Your 1.5M should be under your 2M threshold... you are still getting Low Traffic?